Some proteins were mentioned as good target for drug development since they have a predicted essential function for the parasite.
Despite the lack of extensive information on experimentally verified essential genes, the little evidence that exists, supports our predicted essential roles for a number of genes: metK [ 49], guaA [ 50] and ntpA-B-C-D-E-F-K [ 51].
In addition to the shared predicted essential reactions for all E. coli strains examined, there were two lineage-specific reactions predicted as essential under both aerobic and anaerobic conditions for both EHEC strains (iEco1344_EDL933, iEco1345_Sakai), and the corresponding reactions were for fumarate reductase and glycolate oxidase.
To account for that effect, we used functional homologs to identify potential human cross-reactivity targets for predicted essential C. difficile proteins.
The number of predicted essential genes was fairly similar for the two organisms, with E. coli having 237 (out of 1261) essential metabolic genes and S. typhimurium having 200 (out of 1083) essential metabolic genes.
For predicting essential genes for P. aeruginosa we yielded an area under the curve (AUC) of 0.80 and 0.79 when compared to the experimental datasets paeL and paeJ, respectively.
Moreover, our study suggests that with minimal additional effort, our results can be used to predict essential genes for most bacteria for which an annotated genome sequence is available.
We applied our trained machines from all four datasets (ecoB, ecoG, paeL, and paeJ) to predict essential genes for S. typhimurium and obtained votes from four hundred machines for each gene of S. typhimurium to be essential.
We started predicting essential genes for E. coli.
The trained machines were then applied to predict essential genes for the query organism (E. coli).
We yielded a precision of 67% (accuracy: 87%, sensitivity: 7%, validating with paeL) and 100% (accuracy: 80%, sensitivity: 3%, validating with paeJ) when predicting essential genes for P. aeruginosa.
